Arrhythmia Classification Method Based on Position Convolutional Attention Network

Accurately identifying arrhythmias is critical in electrocardiogram (ECG) research. Although many automatic classification methods have been proposed, it is still a challenge to achieve efficient and accurate classification due to the difficulty of traditional methods in capturing the complex nonlinear and diversity features in ECG signals, the lack of adaptability of the model to different data distributions, and the limitation of feature extraction and model generalization capabilities caused by the lamp problem. To further streamline feature preprocessing and promote data visualization, this study introduces a Position Convolutional Attention Network. The method achieves an average classification accuracy of 97.8% on the MIT-BIH arrhythmia dataset.